404 Prof. J. H, Poynting on 



The air-pressure was reduced as before. When a beam of light 

 from a Nernst lamp was sent through the system, as shown in 

 the figure, it was shifted parallel to itself through a distance 

 about 1'64 cm. The torsion-arm moved round clockwise by 



Fiff. 7.— Plan. 



an easily measurable amount. In one experiment the deflexion 

 was 3*3 scale-divisions, indicating a couple 1*84 X 10~ 5 cm.- 

 dyne. The same beam directed on to the blackened silver 

 disk gave the linear energy-density as 9*8 x 10~ 6 , which should 

 have given a couple 1*6 x 10 -5 . Though the agreement is 

 perhaps accidentally close, yet, as we could use a Nernst lamp, 

 the measurements were much more trustworthy than in the 

 last experiment. 



The interesting point here is that the effect could only be 

 produced by a force outwards at B and E. Whatever forces 

 exist at C and D would be normal to the surfaces and would 

 give no twist. 



A very short experience in attempting to measure these 

 light-forces is sufficient to make one realise their extreme 

 minuteness — a minuteness which appears to put them beyond 

 consideration in terrestrial affairs, though I have tried to show* 

 that they may just come into comparison with radiometer- 

 action on very small dust particles. 



In the Solar system, however, where they have freer play 

 and vast times to work in, their effects may mount up into 

 importance. Yet not on the larger bodies ; for on the earth, 

 assumed to be absorbing, the whole force of the light of the 

 sun is only about a 50 million-millionth of his gravitation- 

 pull. But since the ratio of radiation-pressure to gravitation- 

 pull increases in the same proportion as the radius diminishes 

 if the density is constant, the pressure will balance the pull 

 on a spherical absorbing particle of the density of the earth 

 if its radius is a 50 billionth that of the earth — a little over a 

 hundred-thousandth of a centimetre, say, if its diameter is a 

 hundred-thousandth of an inch. 



* < Nature,' Dec. 29, 1904, p. 200. 



